With the advancement of computer technology and big data, convolutional neural networks of the deep learning have become the mainstream technology for processing large-scale data with grid structure, especially in the field of computer vision. Convolutional neural networks have also been gradually applied in the field of atmospheric science to process multi-angle and multi-scale meteorological data. This paper reviews the progress of convolutional neural networks and their applications in atmospheric science, the conclusions are as following. Through the optimization of network depth and width and magnitude compression, the accuracy and efficiency of convolutional neural networks have been significantly improved, and they have become the mainstream technology for computer vision tasks. The convolutional neural network can process meteorological data efficiently, and has been applied in meteorological target recognition, extreme event detection, numerical model improvement and drought weather event prediction, etc., showing a good application prospect. The application of convolutional neural networks in atmospheric science is still in the exploratory stage, and faces challenges such as the complexity of meteorological data, the need for improvement of model structure and poor interpretability, so further research is needed to promote its development.

The thermal and dynamic processes of the convective boundary layer over the Tibetan Plateau (TP) have an important impact on weather and climate of the downstream region and even the entire East Asia region. This paper uses a summer case of 2017 as an example to analyze the applicability of three sets of reanalysis data including ERA-Interim, JRA-55 and MERRA-2 in the study of the boundary layer over the TP, and further uses the constraints of the numerical model physical framework to correct its analysis error. In the summer of 2017, the variation of air temperature and dew point temperature were presented well through the three sets of reanalysis data in the boundary layer over the southeastern TP, while the reproducibility of the horizontal wind field was very poor, and the reanalysis data with better applicability over the TP during the study period was ERA-Interim. The results from the 12 parameterization scheme combinations selected in this paper were compared by the dispersion degree of the horizontal wind field error, improvements of the simulations in clear skies and moderate rain were significant. Therefore, for the simulated critical physical quantity (the horizontal wind field), the combination of Betts-Miller-Janjic, WSM6 and ACM2 scheme was the most locally applicable in the study area. The wind field in the reanalysis data could describe the summer boundary layer development over the TP more closely after adjustment using simulation results. It was proved that the model parameterization scheme could reduce its deviation of seasonal distribution in the plateau area, which had certain guiding significance for subsequent research and application.

Lanzhou is a typical valley city in Northwest China. Based on the ground conventional meteorological observation data, air quality index at Lanzhou and Baiyin stations and NCEP-DOE reanalysis Ⅱ data, the weather situations during the severe and extreme air pollution processes in Lanzhou from 2015 to 2017 were classified, and the similarities and differences of air pollution between Lanzhou and its satellite city Baiyin under different weather situations were analyzed. And on this basis the influence of meteorological elements on air pollution under different weather situations were discussed by numerical simulation of 21 typical cases. The results are as follows: (1) The weather situations during the severe and extreme air pollution in Lanzhou were mainly trough-ridge pattern and westerly wind pattern from 2015 to 2017. (2) The valley wind circulation was weaker under trough-ridge pattern weather situation. When the degree of air pollution was the same in two cities, the near ground wind field converged over Lanzhou, while the convergence and divergence of wind field wasn’t obvious over Baiyin, and the downward movement over Baiyin was more significant than that over Lanzhou in the afternoon. When the degree of air pollution was different in two cities, the wind field appeared convergence near ground of Lanzhou, and there was a weak downward movement at night, while the wind field appeared divergence near ground of Baiyin, the downward movement wasn’t obvious. (3) The valley wind circulation was stronger under westerly wind pattern weather situation. When the degree of air pollution was the same in two cities, the convergence and divergence of wind field wasn’t obvious over Lanzhou and Baiyin, but there was downward motion near ground of Lanzhou at night. When the degree of air pollution was different, the wind field appeared convergence near ground of Lanzhou, and the downward airflow was dominant in the daytime, while the convergence and divergence wasn’t obvious over Baiyin, and the updraft was dominant near ground during the daytime and nighttime. (4) Under two patterns weather conditions, the wind speed near ground of Lanzhou was usually smaller than that over Baiyin on polluted day, the inversion gradient over Lanzhou was usually larger than that over Baiyin, and the height of boundary layer over Lanzhou was smaller than or close to that over Baiyin.